Initial values #

Initial values determine the starting point of the sampler, and can play a large role in ensuring efficient sampling and convergence. Implementing some degree of randomness is also important when defining initial values to ensure that a model is able to converge properly for a range of initial values. Providing initial values that are within or near the posterior distribution of the parameter of interest serves not only to speed up convergence, but also to optimize the efficiency of the sampler (i.e. make it more random and less autocorrelated) in cases when the sampling algorithm is doing hyperparameter tuning during the early phases of sampling.

Syntax #

# Initial values (one of: 'none', 'default', or 'inherited')
initial values: inherited

Usage #

A value of none leads to JAGS using NULL inits, while default attempts to use a set of relatively functional anonymous inits. inherited uses defaults at first, and then caches posteriors as inits for subsequent fits. In the case of the beta-hurdle model, there is one additional optional value, empirical, that can be specified.

It is also possible to supply a file to inherited values with your own inits.